Condition Assessment of Joints in Steel Truss Bridges Using a Probabilistic Neural Network and Finite Element Model Updating

The condition of joints in steel truss bridges is critical to railway operational safety. available methods for the quantitative assessment different types joint damage are, however, very limited. This paper numerically investigates feasibility using a probabilistic neural network (PNN) and finite element (FE) model updating technique assess bridges. A two-step identification procedure developed achieve localization severity assessment. series FE models with single or multiple damages are simulated generate training testing data samples validate effectiveness proposed approach. influence noise on accuracy also evaluated. results show that change rate modal curvature (CRMC) can be used as damage-sensitive input PNN preliminary exceed 90% when suitable patterns utilized. Damaged members localized correct substructure even contamination. method effectively quantify deterioration robust noise.